This image is a photograph of a poster presentation at the CVPR (Computer Vision and Pattern Recognition) conference held in Seattle, WA from June 17-21, 2024. The poster is titled "Zero-Reference Low-Light Enhancement," authored by Wenjing Wang, Huan Yang, and Jianlong. It details a method for enhancing low-light images using zero-reference learning, which relies solely on normal light images, reducing the need for supervision. The introduction outlines the aim to improve robustness in data usage, illumination-specific hyper-parameters, and adaptation to unseen scenarios. The methodology section describes the design of an illumination-invariant prior to bridge normal light and low-light images, with diagrams illustrating the training process and inference on low-light images. The method section elucidates the training framework, involving physical quadruple prior prediction and image reconstruction. Additionally, the poster shows solutions for detail degradation and showcases the enhancement effects on various test images. The background of the photo includes faintly visible elements of the conference venue and attendees.
Text transcribed from the image:
CVPR
JUNE 17-21, 2024
SEATTLE, WA

Zero-Reference Low-Light Enhancement
Wenjing Wang¹ Huan Yang² Jianlong Yang¹

Introduction
Zero-Reference Low-Light Enhancement: learn solely with normal light images, reducing the need for supervision

Our aim: improve the robustness to
   - Data usage during training
   - Illumination-specific hyper-parameters
   - Unseen scenarios

Our methodology: design an illumination-invariant prior that serves as a bridge between normal light and low-light images

Training on normal light images
Normal Light Input
Physical Quadruple Prior
Illumination Invariant Features
Prior-to-Img Framework
Output
Training framework
   - Predict a physical quadruple prior
   - Reconstruct the prior back to image

Solution of detail degradation: bypass
Low-light enhancement effects for different